Webb finds the strongest evidence yet for ‘black hole stars’ in the early universe

Something unusual inhabits the first billion years of cosmic history, and the James Webb Space Telescope has been revealing it one spectrum at a time. Since Webb began science operations in 2022, it catalogued a puzzling and populous class of objects in the early universe: compact, intensely red sources that appear far too luminous and too small to be ordinary galaxies, far too quiet in X-rays to be conventional active galactic nuclei, and far too point-like to be resolved star-forming halos. Astronomers called them little red dots, or LRDs, and despite four years of sustained inquiry, their physical nature remained contested. Now, the deepest spectrum ever recorded of one of these objects has assembled more than 40 independent lines of evidence into a single coherent picture. The results, published on June 10, 2026, in The Astrophysical Journal, represent the strongest case yet for what little red dots actually are: supermassive black holes growing at extreme rates inside dense cocoons of partially ionized gas, a configuration so distinctive it has earned its own designation — the black hole star, or BH*, model.

Four years of an unresolved puzzle

The term black hole star is not meant to evoke a collapsing stellar object. It refers instead to a physical configuration in which a rapidly accreting supermassive black hole and its surrounding radiation environment are so completely enveloped in a thick shell of gas that the entire system appears point-like and stellar in character, without the extended nebular or galactic structure expected from massive high-redshift objects. From the observer’s perspective, the cocoon reprocesses and scatters the intense radiation produced near the black hole, transforming what might otherwise be a dazzling AGN into something compact, red, and deceptively quiet.

When little red dots first appeared in early Webb data, they posed a diagnostic challenge on multiple fronts simultaneously. Their optical spectra showed broad emission lines — a feature classically associated with the broad-line regions of AGN, where gas orbiting near accreting black holes is accelerated to thousands of kilometers per second. Yet X-ray observatories found them unexpectedly faint. LRDs also lacked the extended light profiles of massive galaxy halos, despite some interpretations requiring enormous stellar masses to account for their luminosities. And they appeared in numbers and at cosmic epochs that certain theoretical frameworks struggled to accommodate without revision. These contradictions held for years, with no single model accounting simultaneously for all observed properties.

GLIMPSE-17775 and nature’s magnifying glass

The object at the center of this new study, designated GLIMPSE-17775, was not originally a target. It appeared serendipitously in data collected by the GLIMPSE programme, a Webb initiative led by Hakim Atek of the Institut d’Astrophysique de Paris. The programme’s primary objective was to search for Population III stars — the universe’s first-generation stellar objects, formed almost entirely from primordial hydrogen and helium before any heavier elements had been produced in stellar interiors — and for extremely faint distant galaxies in the direction of the massive galaxy cluster Abell S1063. Searching near such clusters is strategically sound: their enormous gravitational fields bend and amplify the light of background objects, acting as natural telescopes of exceptional scale.

GLIMPSE-17775 lies behind Abell S1063 at a cosmological redshift of approximately 3.5, meaning we observe it as it appeared roughly 1.8 billion years after the Big Bang. Its position behind the cluster proved decisive. Webb accumulated 30 hours of spectroscopic integration on the source, but the gravitational amplification provided by Abell S1063 expanded the effective information content of that exposure to the equivalent of approximately 80 hours of unaided observation. The result was the most detailed spectrum of a little red dot ever obtained: more than 40 individually resolved spectral lines, each carrying independent information about the physical conditions within the source. A team led by Vasily Kokorev at the University of Texas at Austin recognized what they had. Examining that spectrum, Kokorev described the experience as finding a puzzle scattered across a floor — where, piece by piece, each measurement began revealing a single recognizable image.

The physics of 40 converging lines

Each spectral line in an astronomical spectrum is a diagnostic. It encodes a specific wavelength at which a particular atom or ion emits or absorbs radiation, governed by quantum mechanics and exquisitely sensitive to temperature, density, velocity, and the surrounding radiation field. When 40 lines from different elements and ionic states all independently point toward the same physical scenario, the case for that scenario becomes compelling in a way no single feature, however prominent, can achieve on its own.

The first major diagnostic in GLIMPSE-17775’s spectrum is the shape of its emission lines. Hydrogen, oxygen, and helium all show significantly broader profiles than any simple model of a rotating or turbulent gas cloud can reproduce. The process that best fits these profiles is electron scattering: as photons travel outward through a dense ionized medium, they collide repeatedly with free electrons, each collision imparting a small random velocity component that slightly shifts the photon’s frequency. The cumulative effect of many such scatterings broadens emission line profiles in a characteristic and calculable way. The degree of broadening observed in GLIMPSE-17775 demands an extraordinarily dense, layered ionized cocoon surrounding the source — precisely the structure the BH* model predicts.

The second major diagnostic is what Kokorev’s team has dubbed the iron forest: 16 iron emission lines detected simultaneously in the spectrum. Iron is a heavy element that requires substantial photon energy to strip of its outer electrons, pushing it into highly ionized states. Producing iron at the line strengths and mutual ratios observed in GLIMPSE-17775 requires a radiation field of extreme intensity, one generating sufficient ultraviolet and X-ray photons to drive these high-energy ionic transitions continuously. No realistic stellar population, regardless of mass or age, can sustain such a field. An accreting supermassive black hole generating a hard non-thermal continuum is the natural and essentially unavoidable explanation.

Helium provides two further independent lines of evidence. Fluorescence signatures in helium arise when helium atoms absorb intense short-wavelength radiation and re-emit at characteristic longer wavelengths — a process that requires an intense ultraviolet or X-ray irradiating source. Absorption features, by contrast, require helium atoms to occupy a dense medium between the observer and the radiation source, intercepting part of the continuum along the line of sight. Both signatures appear in GLIMPSE-17775’s spectrum, and each independently reinforces the same physical picture: a dense, layered gas environment enveloping an extreme central engine. Additional oxygen line ratios, whose relative strengths require ionization parameters characteristic of AGN rather than stellar sources, further strengthen the case.

No single one of these features would be conclusive in isolation. Together, in a spectrum of sufficient depth to detect all of them simultaneously from the same object, they constitute an interlocking body of evidence of a quality no previous LRD dataset had been able to provide.

The X-ray silence, finally explained

Among the most counterintuitive properties of little red dots has been their faintness in X-ray observations. Active galactic nuclei are among the most luminous X-ray emitters in the universe. Their accretion disks and hot coronae produce copious high-energy radiation detectable across cosmological distances. If LRDs harbor rapidly accreting black holes, they ought to be prominent in X-ray surveys. Yet observationally, they are not.

The BH* model resolves this contradiction without invoking special assumptions about intrinsic luminosity. The dense gas cocoon that reprocesses optical and infrared photons is simultaneously optically thick to X-ray radiation. The column density of ionized gas along the line of sight is high enough to absorb X-ray photons before they escape the cocoon. The central engine may be radiating energetically across a broad spectrum, but the X-ray component is intercepted and thermalized within the surrounding gas shell. What reaches the observer is the reprocessed, longer-wavelength emission that makes these objects appear red and point-like. The electron scattering signatures in GLIMPSE-17775’s spectrum, and the inferred gas densities they imply, are consistent with column densities sufficient to produce exactly this level of X-ray suppression. The quiet X-ray profile of LRDs is not evidence against an AGN interpretation — it is evidence for extreme obscuration of one.

Cosmology is not broken

When little red dots first accumulated in Webb’s early datasets, a portion of the astrophysical community raised a pointed concern. Under the interpretation that LRD luminosities arose primarily from stellar populations, the implied stellar masses seemed difficult or impossible to reconcile with what cosmological simulations predict for galaxy formation in the first billion years of cosmic history. The concern was expressed informally as LRDs «breaking cosmology» — requiring either radical revision of our galaxy formation models or some undiscovered physical process.

The BH* interpretation eliminates this tension by removing its premise. If the luminosity comes predominantly from accretion onto a black hole, mediated and reprocessed by the surrounding cocoon, the implied stellar masses are far more modest. The broad emission lines are produced by the high velocities of infalling gas near the black hole, not by the total mass of stars in the galaxy. The black hole masses required under this scenario are lower than those estimated under purely stellar models, and the overall energy budget fits comfortably within standard cosmological simulations.

GLIMPSE-17775 adds a further element. Combined Webb and archival Hubble Space Telescope data — drawn from the Frontier Fields and BUFFALO programmes — revealed a substantial surrounding host galaxy contributing extra blue-wavelength light to the source. This explains why GLIMPSE-17775’s Balmer break, a spectral dip that arises when stellar continuum is superimposed on the AGN spectrum and is considered a hallmark feature of LRDs, appears shallower than in other class members. A significant host galaxy dilutes this signature, and the BH* model explicitly attributes excess blue light to the host stellar population. The object slots into the existing cosmological framework without friction, and the consistency with standard structure formation models is, as Kokorev summarized in the announcement, something that makes «the puzzle that is our universe even better.»

What comes next

The GLIMPSE-17775 result is as much a methodological benchmark as a scientific conclusion. It demonstrates that combining Webb’s infrared spectroscopic depth with the natural amplification of foreground gravitational lensing can yield LRD spectra rich enough to discriminate rigorously between competing physical models. The threshold it sets is demanding: not two or three diagnostic features, but more than 40, all measured independently from the same object, all consistent with a single picture.

Kokorev’s team has been explicit about what remains open. The precise mass of the central black hole, its accretion rate relative to the Eddington limit, the gas properties of the cocoon at high spatial resolution, and whether the BH* scenario applies across the full LRD population or predominantly to the most extreme examples are all questions that require additional data. GLIMPSE-17775’s unusually prominent host galaxy suggests it may differ in some respects from the broader LRD sample, and population-level generalizations await more spectral observations of comparable depth.

Alternative theoretical proposals involving exotic central energy sources with properties intermediate between ordinary stars and supermassive black holes continue to be developed, and the fundamental question of what powers the central engines of these objects — while now substantially constrained — is not fully closed. What this paper establishes, with greater observational depth than any predecessor, is that at least one little red dot behaves in precise accordance with BH* model predictions across 40 independent spectral diagnostics simultaneously. For a field that has been assembling this puzzle piece by piece since 2022, that is a significant threshold. The full mosaic is not yet complete. But the dominant shapes within it are now visible.

The paper was led by Vasily Kokorev (University of Texas at Austin) and published on June 10, 2026, in The Astrophysical Journal. DOI: 10.3847/1538-4357/ae4ed7. Principal Investigator of the GLIMPSE programme: Hakim Atek (Institut d’Astrophysique de Paris). Supplementary data from the Hubble Space Telescope Frontier Fields and BUFFALO programmes. Image credit: NASA, ESA, CSA, Vasily Kokorev (UT Austin), Alyssa Pagan (STScI).

If you value rigorous, independent astrophysics journalism from Costa Rica — the kind that goes beyond the press release — share this article and follow SKYCR.ORG.

© 2026 SKYCR.ORG | Homer Dávila Gutiérrez, FRAS. All rights reserved. Total or partial reproduction is prohibited without express authorization. Original source: https://esawebb.org/news/weic2610/